| Autor: |
Siouda, Roguia, Nemissi, Mohamed, Seridi, Hamid |
| Zdroj: |
Evolving Systems; Feb2023, Vol. 14 Issue 1, p37-48, 12p |
| Abstrakt: |
This paper introduces a heartbeat classification system that combines three types of neural networks: random neural networks, deep autoencoders and RBF neural networks. The aim is to make use of the advantages of these neural networks in order to introduce a model with simpler architecture than the state-of-the-art deep models. Indeed, the advantages of the three combined networks, briefly, are these: (i) Autoencoders provide high level features without pre-processing; (ii) Random neural networks provide good generalisation and very fast training; (iii) RBF neural networks provide high coverage of the input space and allow using prior knowledge. On the other hand, two types of features are used: coded features (obtained from the autoencoder) and RR interval based-features. To evaluate the performance of the proposed system, we conduct experiments on the MIT-BIH arrhythmia dataset and we consider the recommendations of the association for the advancement of medical instrumentation, which defines five classes of interest. Furthermore, the experiments are based on an inter-patient paradigm and the obtained results are compared with some of the state-of-the-art methods. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink